Constitutive artificial neural networks: A fast and general approach to predictive data-driven constitutive modeling by deep learning

Abstract In this paper we introduce constitutive artificial neural networks (CANNs), a novel machine learning architecture for data-driven modeling of the mechanical behavior materials. CANNs are able to incorporate by their very design information from three different sources, namely stress-strain data, theoretical knowledge materials theory, and diverse additional (e.g., about microstructure or processing). can easily efficiently be implemented in standard computational software. They require only low-to-moderate amount training data time learn without human guidance also complex nonlinear anisotropic Moreover, simple academic example demonstrate how input microstructural endow with ability describe not known but predict properties new where no available yet. This may particularly useful future in-silico The developed source code CANN accompanying sets at https://github.com/ConstitutiveANN/CANN .